You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
You'll learn how to tune NGINX on Arm-based platforms using a measurement-first
36
+
workflow. First, you'll establish a baseline, then iteratively adjust Linux network stack parameters and
37
+
evaluate compiler and library choices such as OpenSSL, PCRE, and zlib. You'll review tuned configuration examples and guidance for sizing directives to refine configurations
38
+
for static file serving, reverse proxy, and API gateway scenarios. You're also introduced to `wrk`
39
+
as an option for generating traffic and reporting throughput and latency on NGINX. By the end, you'll use `wrk` to compare before-and-after results and verify the impact of each change.
40
+
faqs:
41
+
- question: How do I know whether a tuning change improved performance?
42
+
answer: >-
43
+
Re-run the same repeatable workload and compare throughput and latency before and after
44
+
the change. Adjust one parameter at a time or use a designed experiment so results are attributable
45
+
to specific settings.
46
+
- question: Where do I change Linux network parameters when tuning?
47
+
answer: >-
48
+
Modify settings in /etc/sysctl.conf or use the sysctl command. Refer to the Linux kernel
49
+
sysctl documentation, and treat these changes as part of the same measurement process used
50
+
for NGINX directives.
51
+
- question: Do I need to use wrk if I already have a load test?
52
+
answer: >-
53
+
No. If you already have a repeatable HTTP workload, use that. wrk is provided as a starting point
54
+
only when you do not already have an established test method.
55
+
- question: When tuning the reverse proxy or API gateway, what should I reuse from the static
56
+
file server configuration?
57
+
answer: >-
58
+
Use the same top-level nginx.conf from the static file server section. Then apply the tuned
59
+
reverse proxy and API gateway configuration and validate the result with repeatable measurements.
60
+
- question: How do compiler, OpenSSL, PCRE, and zlib choices factor into tuning?
61
+
answer: >-
62
+
Evaluate these choices as part of your experiments because they can affect NGINX performance.
63
+
Compare alternatives using the same workload and interpret results alongside your NGINX
64
+
and kernel changes.
65
+
# END generated_summary_faq
66
+
20
67
author: Julio Suarez
21
68
22
-
generate_summary_faq: true
69
+
generate_summary_faq: false
23
70
rerun_summary: false
24
71
rerun_faqs: false
25
72
@@ -73,3 +120,4 @@ weight: 1 # _index.md always has weight of 1 to order corr
73
120
layout: "learningpathall"# All files under learning paths have this same wrapper
74
121
learning_path_main_page: "yes"# This should be surfaced when looking for related content. Only set for _index.md of learning path content.
0 commit comments